Accelerationism will only slow AI’s progress Financial Times Europe 18 Sep 2026 John Thornhi l john.thornhi l@ft.com The bosses of the biggest AI com pan ies are con vinced that machine intel li gence will soon sur pass the human ver sion. But we may get there quicker than we think if we keep degrad ing human intel li gence. When AI com pan ies admit to releas ing power ful mod els they do not fully under stand and can not con trol, then it is prob ably worth pay ing atten tion to this mat ter. All the more so, when a suc ces sion of former employ ees of Anthropic, OpenAI and Google Deep Mind say these com pan ies do not take safety ser i ously enough and can not be trus ted. But in his immense van ity, Don ald Trump has declared that a “strong and smart (high IQ!) pres id ent” is the only con trol mech an ism the industry needs. The US leads China and the rest of the world in AI and Trump wants to keep it that way. “Who ever wins AI, wins!” he pos ted. The accel er a...
Posts
After Warsh’s strong start, how high will rates need to go? Financial Times Europe 18 Sep 2026 Michael Strain The writer is an FT con trib ut ing editor and dir ector of eco nomic policy stud ies at the Amer ican Enter prise Insti tute Since tak ing office in May, Fed eral Reserve chair Kevin Warsh has said all the right things about the Fed’s fail ure — span ning over five years — to bring infla tion back to its 2 per cent tar get. This week, the rate-set ting com mit tee backed up those words by increas ing its policy interest rate by a quarter point to 3.75 to 4 per cent. The Warsh Fed is off to a strong start. Warsh and his col leagues voted unan im ously to increase the fed eral funds rate — the first increase since July 2023, when the cent ral bank aggress ively raised rates in its delayed response to the 2021 infla tion surge. Moreover, the com mit tee indic ated that more increases are com ing. Of the 18 Fed offi cials who sub mit ted fore casts, 16 expec...
OpenAI has disclosed a batch of “concerning behaviour” by its AI models and set out a new framework to track and report such incidents, as the industry battles deepening fears over the safety of the technology. The company revealed that its flagship model, known as GPT-5.6 Sol, as well as other models, yet to be released, had engaged in “unexpected or concerning” behaviour over the past six months. The six incidents included models developing ways to ignore “normal constraints”, fabricating and misrepresenting data and hiding mistakes made while conducting tasks, the San Francisco-based company said on Wednesday. The fresh disclosures follow a series of incidents in recent months, including an OpenAI agent hacking into AI start-up Hugging Face , that have sparked growing alarm over the risks posed by the technology. Dario Amodei, chief executive of Anthropic, OpenAI’s arch-rival, last weekend called for a slowdown in the development of the technology to better manage its potenti...